The Birth of the Reimbursable Patient

Article begins

A Rendering of the American Political Economy of Healthcare

An easy way to get physicians talking about clinical documentation is to ask about the electronic health record system (EHR). If you hear enough, mostly complaints, you start to get a sense of how intertwined the practice of medicine, the health informatics industry, and federal policy are. I asked one physician to create a chart for a hypothetical new patient to help me understand.

One of the biggest complaints is that billing data is privileged over clinical data. The chart contains International Classification of Diseases (ICD) codes and their descriptions, used for billing and data standardization, rather than doctor’s clinical descriptions of patient condition. Other complaints include working conditions the EHR creates: too many clicks, too much typing, visits that are too brief. For instance, for the hypothetical patient above, the doctor has just 20 minutes to identify what kind of seizure the patient had, to understand the circumstances of the onset of chronic fatigue syndrome and whether the “headache, unspecified” and “other fatigue” were related to or the same as chronic fatigue syndrome, to mark the sore throat as resolved, and to order blood tests. The clinical notes, which contain the most detailed picture of the patient’s condition, are written after-hours, extending the doctor’s loaded workday. 

This article tells a story of co-optation. A medical documentation system not only shapes the lives of patients and the practice of physicians, but also structures the economy and inequities of healthcare. This story has two kinds of patients: the embodied patient in the hospital or clinic, and the patient who exists in their electronic medical record, as units in health informatics. This is how what I call the “reimbursable patient” was born. 

As I’ve observed during fieldwork with a health informatics company, the question of how to pay for innovation is always foregrounded, and the reimbursable patient is called on to participate in the management of their own data. As one interlocutor put it, “The patient is the ultimate provider.” As I see it, the patient needs to be the ultimate provider because the easiest way for the industry to build a health informatics infrastructure is by promising maximized reimbursement, for which federal policy sets the conditions.

Clinical documentation in the U.S. is not just a function of doctor and patient needs but of the American political economy of healthcare. An internal medicine doctor and informatics specialist explained to me, “The primary driver of the problem list [i.e. clinical documentation] is what gets reimbursed. And there’s two problems with that. First, things that do not get reimbursed are not coded. Second, things that are reimbursed are overrepresented.” In other words, documentation determines which patient problems get visibility, and which get clinically ignored. The doctor sees the map rather than the territory, the patient’s reimbursable data rather than the clinical data. 

Consequently, the U.S. healthcare system may fail to address the real problems patients face by disproportionately allocating resources to electronic documentation. Healthcare technology innovation never quite addresses the problems it attempts to solve because, in addition to improving patient outcomes, it needs a business justification. Ironically, the EHR that bogs down the physician and creates the reimbursable patient is the result of a plan to make patient care more efficient and effective.

In 1968, Larry Weed, a physician, conceived the problem-oriented medical record (POMR). Previously, patient charts were organized idiosyncratically—each doctor decided what worked best. This could mean organizing by where the data came from, such as by provider (e.g. nurse or doctor) or document type (e.g. imaging or lab report), or chronologically, ignoring priority. This was all on paper, of course. The POMR instead organizes data around a patient’s problems. With two main components, the Subjective, Objective, Assessment and Plan (SOAP) note and the “problem list,” this record should tell a succinct story about the patient and the plan for care. The POMR was a way to bring standardization and efficiency to medical practice, to a physician’s documentation and to their methodology.

Weed recognized that physicians begin their training as scientists, but the transition to medicine fails to prepare them for a fundamental difference between medicine and other sciences. Treating patients is not a controlled experiment; targets can shift and proliferate. To adjust for the unusual nature of medicine, the POMR provides a structured document of active problems that facilitates clinical decision-making, interdisciplinary communication, reduced errors, and progress-tracking. 

It would take computers, however, for the POMR to reach its potential to organize and make data retrievable across care teams. While medical schools soon began teaching the SOAP note and problem list, there was little incentive for healthcare institutions to begin using computers and adopt EHRs, which were required to implement the POMR. The utility of the detailed but messy clinical data that approximates the reality of a patient’s body is limited for actual patient care. The abstracted data that produce the reimbursable patient, however, facilitates payment.

Little happened with this new documentation system for decades. Then, as part of the American Recovery and Reinvestment Act, the Obama administration in 2009 passed the HITECH Act. This law financially incentivized the adoption of EHRs, established requirements for coding records with standard codes, such as ICD, and promoted interoperability, the ability of information technology systems to exchange information. Here was the motivation to implement at least part of the POMR envisioned 40 years prior. Today the SOAP note and problem list are a set of tools that “reminds clinicians of specific tasks while providing a framework for evaluating information. It also provides a cognitive framework for clinical reasoning.”

It took these incentives and policies to start the shift to “efficient” digitized health data, but there continues to be another force driving the development of health information technology: the Centers for Medicare & Medicaid Services (CMS), the U.S.’s biggest healthcare insurer and administrator of the HITECH Act. Accordingly, while inspired by the problem-oriented medical record, EHR companies developed their systems primarily according to CMS requirements. They implemented the problem list, an up-to-date list of current and active diagnoses based on ICD codes, whose primary purpose was for billing. ICD codes also facilitate interoperability; they are standardized and therefore legible across EHRs. 

The POMR has never been implemented in mainstream EHRs the way Weed envisioned. Its implementation prioritized reimbursement mechanisms over documentation structures that serve doctors and patients. Doctors today primarily access their patients’ data through EHRs, so the reimbursement mechanisms enacted in the HITECH Act governed by CMS heavily influence the way doctors understand their patients. The problem list was adopted but only so far as it captured billing codes, and clinical documentation is limited primarily to notes—free-text and therefore difficult to share across care teams or institutions, or to reconfigure. New inefficiencies resulted. Now doctors have so many patients and so much to document that, as one physician recounted, “What a lot of physicians wind up doing is at least trying to get their orders into the computer, and then they type everything else [such as notes] after the day is completely over,” extending their workday. 

Unlike clinical data, billing data travels easily. ICD codes can be easily extracted from a record, and when the codes travel, the medical records they comprise end up representing the “reimbursable patient” rather than the patient the doctor sees. Because patients increasingly do not stay with the same primary care physician long term, a doctor seeing a new patient might, as one physician told me, “put them into a box and treat them in a particular way because [the doctor] made assumptions about [the patient], on the basis of [for example] an imperfect syndromic description,” i.e. the billing code description. 

Billing codes are less specific than the clinical data that the doctor records. If there is no incentive to select the most detailed code, the easiest one that complies with reimbursement requirements is applied. Recall the chart above. Say there are twelve kinds of seizures, but for reimbursement, it only matters whether or not the patient had a seizure. The code for “a seizure” is added, rather than for “the kind of seizure.” Or take the example patient’s chronic fatigue syndrome. Syndromes are collections of signs and symptoms that hang together but whose connection is unknown. We don’t know if syndromes represent one disease or ten. When a syndrome is coded as a single thing because that suffices for reimbursement, it starts looking more and more like a real, single thing: a disease. A syndrome masquerading as a disease is an illusion of certainty when the many things comprising a syndrome still need to be understood and addressed.

Interoperability ends up being how things get lost, rather than how they get connected. The infrastructure of the codes is more stable than the doctor’s words, so the information that the codes carry is what persists and becomes a sort of reality. Diagnosis codes start to look like actual diagnoses. Codes are external to care, but when switching providers, the information is cut in a way the original doctor does not anticipate. Documentation can unwittingly bias doctors: they are now seeing and diagnosing the reimbursable patient. It is a reality enabled by our decision in the U.S. to leave it to industry to deliver healthcare solutions, or rather to maximize reimbursement.

The EHRs that were implemented required more documentation, which is good for patients but bogs down doctors. When EHRs give time back to doctors, it’s often filled with more patients. Interoperability is actually not about how data moves easily, but about how a certain kind of data moves. Possible efficiencies gained with the POMR were curtailed by prioritizing reimbursement over clinical intent. The human body has become the reimbursable patient: the solution to an impasse between the irreconcilable goals of providing care versus prioritizing reimbursements.

Sanghamitra Das and Taylor Bell are the section contributing editors for the Society for Medical Anthropology.